WO1984000659A1 - Reduced data rate signal separation system - Google Patents

Abstract

A signal separation system operating at a reduced data rate, which requires comparably fewer storage locations than previous arrangements. Such a signal separation system, for use in a television receiver, separates two sampled data video signal components which are interleaved in frequency. This system comprises means (12) for passing the sampled data video signals at a data rate less than the original sampling rate. A comb filter (20) is responsive to the reduced sample rate signal, and produces the comb-filtered output signal.

Description

REDUCED DATA RATE SIGNAL SEPARATION SYSTEM

This invention relates to signal separation systems and, in particular, to a comb filter arrangement for separating the luminance and chrominance components of a digitized video signal at a reduced data rate.

Conventional television broadcast systems are arranged so that much of the brightness (luminance) information contained in an image ,is represented by signal frequencies which are concentrated about integer multiples of the horizontal line scanning frequency. Color

(chrominance) information is encoded and inserted in a portion of the luminance signal spectrum around frequencies which lie halfway between the multiples of the line scanning frequency (i.e., at odd multiples of one-half the line scanning frequency).

Chrominance and luminance information can be separated by appropriately combing the composite signal spectrum. Known combing arrangements take advantage of the fact that the odd multiple relationship between chrominance signal components and half the line scanning frequency causes the chrominance signal components for corresponding image areas on successive lines to be 180° out of phase with each other. Luminance signal components for corresponding image areas on successive lines are substantially in phase with each other.

In a comb filter system, one or more replicas of the composite image-representative signal are produced which are time delayed from each other by at least one horizontal line scanning interval (a so-called one-H delay). The signals from one line are added to signals from a preceding line, resulting in the cancellation of. the chrominance components, while reinforcing the luminance components. By subtracting the signals of two successive lines (e.g., by inverting the signals of one line and then combining the two), the luminance components are cancelled while the chrominance components are reinforced. Thus, the luminance and chrominance signals

OMPI WIPO ay be mutually combed and thereby may be separated advantageously.

The composite video signal may be comb filtered in an analog form, a sampled data form, or a digital form. Comb filters using analog signal delay lines for the

(approximately) one-H delay lines are commonly employed in PAL-type receivers to separate the red and blue color difference signals^', taking advantage of the one-quarter line frequency offset of the interlacing of the two signals. An example of a comb filter system for a sampled data signal is shown in U.S. Patent 4,096,516, in which the delay line comprises a 682*$ stage charge-coupled device (CCD) delay line which shifts signal samples from stage to stage at a 10.7 MHz rate to achieve a one-H delay. The article "Digital Television Image Enhancement" by John P. Rossi, published in Volume 84 of the Journal of the Society of Motion Picture and Television Engineers (1974) beginning at page 37 shows a digital comb filter in which the one-H delay is provided by a digital storage medium for 682 samples which is accessed at a 10.7 MHz rate.

In the CCD delay line described in the above-referenced U.S. patent, 682- stages are needed to transfer charge packets related to the analog video signal. But in the digital delay line described in the Rossi article, the video signal is in the form of eight-bit digital samples. This arrangement requires the use of eight storage locations for each of the 682 samples in a horizontal line, or a storage medium for 5,456 bits. Moreover, this delay line is only of sufficient size for a system in which an NTSC color video signal is sampled at a rate of three times per subcarrier cycle (i.e., using a 10.738635 MHz sampling signal). Another proposed sampling frequency for digitizing the analog video signal is 14.3181818 MHz, or four times the color subcarrier frequency. A one-H digital delay line operating at this frequency requires storage for 910 samples which, at eight bits per sample, requires a total of 7280 storage

OMPI locations. Since a storage medium of this capacity is difficult to fabricate economically, it is desirable to provide a digital comb filter system which requires fewer storage locations. In accordance with an aspect of the present invention, a signal separation system for use in a television receiver separates two sampled data video signal components which are interleaved in frequency. Such a system comprises means for passing the sampled data video signals at a data rate less than the original sampling rate. A comb filter is responsive to the reduced sample rate signal and produces the comb-filtered output signal.

In accordance with a further aspect of the present invention, a signal separation system operating at a reduced data rate is provided, which requires comparably fewer storage locations. A video signal having frequency interleaved signal components is sampled at a given data rate and is applied to a first bandpass filter, which produces a filtered signal restricted to a portion of the video passband of the composite video signal. Sample, are applied to a comb filter operating at a data rate less than the original data rate, to produce a comb-filtered signal. The comb-filtered signal is then applied to a second bandpass filter, which provides a sequence of samples occupying a portion of the video passband. The comb filter arrangement operating at a reduced data rate requires comparably fewer storage locations than previous arrangements.

In one embodiment, a digitized composite video signal of a given sampling rate is applied to a bandpass filter, which produces a filtered signal restricted to a portion of the passband of the composite video signal. The filtered signal is then subsampled at a rate which satisfies the Nyquist criterion for information of the restricted passband. Samples, now at a reduced data rate, are applied to a one-H delay line, and delayed and undelayed signals are combined to produce a first comb-filtered signal. The first comb-filtered signal is then applied to an interpolator, which provides a sequence of samples at the sampling rate of the original digitized composite video signal. This sequence of samples is then combined with the samples of the composite video signal to produce a second comb-filtered signal.

In the drawings:

FIGURE 1 illustrates in block diagram form a digital signal separation system constructed in accordance with the principles of the present invention;

FIGURES 2 and 3 illustrates in block diagram form bandpass filters suitable for use in the signal separation system of FIGURE 1; FIGURE 4 illustrates in block diagram form a digital comb filter suitable for use in the arrangement of FIGURE 1;

FIGURE 5 illustrates in block and logic diagram form an interpolator and a delay circuit suitable for use in the arrangement of FIGURE 1;

FIGURE 6 is a timing diagram showing the instantaneous signal contents at points in the arrangements of FIGURES 2, 4 and 5;

FIGURES 7 and 9 illustrate in block and logic diagram form a clock signal generating arrangement suitable for use with the arrangements of FIGURES 2, 4 and 5;

FIGURES 8 and 10 are timing diagrams showing waveforms illustrating the operation of the clock signal generating arrangement of FIGURES 7 and 9;

FIGURE 11 illustrates in block diagram form a digital signal separation system constructed in accordance with the principles of the present invention;

FIGURE 12 illustrates in block diagram form a bandpass filter and comb filter arrangement suitable for use in the embodiment of FIGURE 11; FIGURES 13 illustrates in block and logic diagram form a clock generator circuit for the comb filter system of FIGURES 12 and 14;

FIGURE 14 illustrates in block diagram form an interpolating filter suitable for use in the embodiment of FIGURE 11;

FIGURE 15 is a timing diagram used to explain the operation of the clock generator circuit of FIGURE 13;

FIGURES 16 and 17 are timing diagrams which show the instantaneous signal contents at points in the embodiments of FIGURES 12, 13 and 14;

FIGURE 18 illustrates in block diagram form another embodiment of a digital comb filter system constructed in accordance with the principles of the present invention;

FIGURE 19 illustrates in block diagram form a digital lowpass filter suitable for use in the embodiment of FIGURE 18; and

FIGURE 20 is a graph showing frequency response characteristics illustrating principles of operation of the embodiments of the present invention.

Referring to FIGURE 1, a signal separation system constructed in accordance with the principles of the present invention is shown. In FIGURE 1, single-line arrows represent paths for individual signals, and broad arrows represent data paths for multibit digital samples.

In FIGURE 1, a composite analog video signal is applied to the input of an analog-to-digital (A/D) converter 10. The A/D converter samples the analog video signal in response to a sampling signal (4F ) and produces digital video samples at the sampling signal frequency (4f ). As used herein, clock signals are designated using upper-case letters (e.g. 4F ) and the frequency or sample rate of signals are designated using lower-case letters (e.g. 4fsc) .

In the embodiment of FIGURE 1, the sampling signal frequency 4fh3C is equal to four times the color subcarrier frequency, where f is the color subcarrier

Is

O.^yf

_/ , ~ frequency. In the NTSC color television system in which the color subcarrier frequency is 3.579545 MHz, the 4f sampling signal frequency is 14.31818 MHz.

The 4f rate digital samples produced by the A/D converter 10 are applied to inputs of a delay stage 14 and a digital bandpass filter 12. The passband of the bandpass filter 12 includes the chrominance frequencies which are distributed about the 3.58 MHz color subcarrier frequency in the NTSC system. The delay stage 14 provides an equalizing delay which matches the delays encountered by signals between the input of the bandpass filter 12 and the output of an interpolator 16.

The bandpass filter 12 produces filtered output signals at a clock rate which is lower than that of the input signals to the filter. This may be accomplished, for example, by subsampling the output signals produced by the filter. In the embodiment of FIGURE 1, the bandpass filter 12 produces output signals at a 2f rate, which is half the rate of the input signals to the filter. The 2f rate of the filtered signals satisfies the Nyquist sampling criterion for the bandwidth of the information-bering filtered signals.

The 2f rate filtered signals are applied to a comb filter 20. The comb filter 20 includes a one-H delay line 22, also clocked at the 2f signal rate, and a subtractor 24. The filtered signals passed by the one-H delay line are applied to one input of the subtractor 24, and undelayed filtered signals are applied to the second input of the subtractor. The sμbtractor 24 will therefore produce comb filtered chrominance signals at a 2f ■S^* clock rate. The comb filtered chrominance signals are now in a form suitable for further processing and display.

The comb filtered chrominance signals are also applied to the input of an interpolator 16. The interpolator 16 interpolates the 2f rate chrominance samples to produce signal values intermediate the 2f samples. The output signal of the interpolator thereby comprises combed chrominance signal samples at the 4f sampling rate. These 4f rate combed chrominance signals are applied to one input of a subtractor 18, which also receives 4f rate samples passed by the delay stage 14. The subtractor 18 subtracts the comb filtered chrominance samples from the composite signal samples passed by the delay stage 14 to produce comb filtered luminance signal samples. The comb filtered luminance signal is available at the output of the subtractor 18 for subsequent luminance information processing. The arrangement of FIGURE 1 is advantageous in that the comb filter 20 operates at a lower data rate than the sampling rate of the composite signal produced by the A/D converter 10. This means that the one-H delay of the delay line 22 can be provided by a 455 stage delay line for NTSC signals, instead of the 910 stage delay line required for a 4f data* rate. For example, if the comb filter produces eight bit filtered samples, only 3640 storage locations are needed in the one-H delay line instead of the 7280 locations required for a 4f comb filter. The 2fsc rate comb filtered chrominance signals are then interpolated to supply chrominance samples at the composite signal rate of 4f and subtracted from the composite signal samples to produce a comb filtered luminance signal. A digital finite impulse response (FIR) bandpass filter suitable for use as bandpass filter 12 in the embodiment of FIGURE 1 is shown in detail in FIGURE 2. In FIGURES 2, 4, 5, 11 and 20 broad lines depict data paths for multibit samples and thin lines depict clock signal paths.

Composite video signals produced by the A/D converter 10 are applied to the input of the first stage of an eight-stage shift register 30, including delay stages τ, through τ_ft. Each delay stage is clocked by a 4F £__ - clock signal so as to store a multibit sample for one cycle of the 4F clock signal. For instance, if the samples are eight bits in length, each delay stage will contain eight bits of information. The video signal

OMPI samples are clocked sequentially through the register 30, with output signals from the last stage τ_Q being applied to a delay stage τ_g in the delay stage 14 of FIGURES 1 and 5. In addition, the input of the first stage τ, and the outputs of the remaining stages are coupled to inputs of data latches of a latch register 40.

The latch register 40 includes nine L, latches labelled L,, through L_ιg, and nine L_Q latches labelled L--, through L_g. The L, and L_Q latches are arranged in pairs at the outputs of the shift register stages, with latches L_T1 and L_Q, having their inputs coupled to the input of shift^* register stage τ,, latches L,₂ and L_Q2 having their inputs coupled to the input of shift register stage t_, and so forth. The L.. latches are loaded in parallel by a clock signal I-.. , and the L_ latches are loaded in parallel by a second clock signal Q _L- The I_CL and Q_CL clock signals in this embodiment are each of the same frequency, f , but are of different phases.

The video signal samples held in the latches 40 are applied to inputs of weighting function circuits 60 by a multiplexer 50. The multiplexer 50, representatively shown in FIGURE 2 as an array of switches, alternately couples the outputs of the L_ and L_Q latches to inputs of the weighting function circuits 60. The multiplexer 50 is switched by an I/ CLOCK„ signal at a 2f rate. The weighting function circuits weight the applied signals by respective weights of 1/8, -7/64, -15/64, 3/64 and 5/16 as shown in the FIGURE. The weighting function circuits may be constructed using shift matrices and adders as described in United States patent application number

363,827 entitled "Digital Filter Circuits", filed March 31, 1982 by Lauren A. Christopher.

Weighted signal samples produced by the weighting function circuits 60 are applied to an adder tree 70, which sums all of the weighted samples at the output of the last adder 78 of the adder tree. The output signal of each weighting function circuit except for the center circuit 62 passes through four adders in the tree j

CMPI to reach the output of adder 78. The output of the center circuit 62 passes through only one adder 78, and thus will normally arrive at the output of adder 78 ahead of the other samples of a corresponding point in time. In order to bring the center circuit samples into the same time relationship as the other samples, a delay τ_3A equal to the delay of three adder tree adders is inserted in the signal path at the output of the center weighting function circuit. The weighted samples will thereby be summed in the proper time relationship at the output of the final adder 78. The summed signal samples at the output of adder 78 are then latched into an output latch L_∑ by the

I/Q CLOCKp, which produces filtered signals I-, Q_ at the

The operation of the bandpass filter of FIGURE 2 is explained with reference to FIGURE 6, which depicts the sequence of operation of the filter. FIGURE 6a illustrates the contents of the shift register 30. Composite video signal samples, identified as I. , Q, , -I-, / -Qm , etc. are shifted through the register 30 at a 4f rate. The arrows in FIGURE 6a indicate the times at which the register is clocked to shift the data. After four shifts, the first sample I_ has been clocked into the t₄ stage. At that time, t, , succeeding samples Q, , -I, , and -Q₁ are located in shift register stages t₃, t₂ and τ, , respectively. Preceding signal samples are stored in stages τ D_c through τ_Q- One-half cycle of the 4FΞ,,,C, clock later, at time t., ₅, the contents of the shift register 30 is loaded into the L.. latches by the l__ clock. FIGURE 6b illustrates the contents of latch L-_rc, which now holds sample I, . It follows that latch L,₄ now holds sample , , latch L_T3 holds sample -I, , and so forth.

At time t₃ , the shift register 30 is again clocked so that sample Q, is now located in stage τ^. One-half cycle of the 4F clock later, at time t_ _ς, the samples held in the shift register are loaded into the L_ latches. FIGURE 6c illustrates the contents of latch _Q5, which contains sample Q, following time t_{3 5} The shift register 30 and latches 40 continue to be clocked in this manner. After three 4F clock cycles t-., t_g and t-, sample I₂ is located in stage τ₄ and the L, latches are reloaded at time t_ ₅- The shift register is shifted again at time t_g and the L_Q latches are loaded at time tg ₅

The multiplexer 50 directs the samples from the latches 40 to the weighting function circuits at a 2f clock rate, as shown in FIGURE 6d. From time t₃ to time t_g the L_χ latches are connected to the inputs of the weighting function circuits by the multiplexer. From time t_c6 to time t_n9 the LQ- latches are connected to the weighting function circuits. The multiplexer continues to alternate between the latches at this 2fsc„ rate. The samples now propagate through the weighting function circuits 60 and the adder tree 70 to the output of adder 78. Since adders can be expected to exhibit appreciable propagation delay as compared to other circuit elements, such delay must be taken into consideration in the operation of the bandpass filter. For purposes of illustration, it will be assumed that the weighting function circuits exhibit propagation delays of 66 nanoseconds, and that each adder exhibits a propagation delay of 40 nanoseconds. Thus, each signal experiences a delay of 226 nanoseconds as it passes from the output of the multiplexer 50 through a weighting function circuit and four adder delays. For instance, signals held in latches L_ are coupled through the multiplexer at time t₃, but do not arrive at the output of adder 78 for 226 nanoseconds, which is a time preceding time t_g by approximately 54 nanoseconds. Therefore, by time t_g the signal at the output of adder 78 has had 54 nanoseconds to settle, and is then clocked into latch L^ at time t_ . It is seen that five cycles of the 4fsc shift register clock (^tι~^"i*₉ *^ⁿ FIGURE 6a) elapse before the filtered signal sample I-.-, corresponding to sample I_. is produced at the output of the bandpass filter, as shown in FIGURE 6e. This delay time is taken into consideration in the selection of the equalizing delay 14 in FIGURES 1 and 5.

An alternate bandpass filter design 12' is shown in FIGURE 3. In this arrangement, the filter delay stages as well as the computational elements are operated at a reduced rate. It will be appreciated that structurally differing FIR filters will produce similar FIR filter functions provided they operate to produce the same sum of weighted samples in the same time sequence. The circuitry necessary to produce this result requires apparatus to successively store applied signal samples and route the requisite samples to the coefficient multipliers in the necessary sequence.

The FIGURE 3 filter circuitry accomplishes this result by applying the input signal samples in an alternating mode to two separate delay registers P and M which operate at half the sample rate. The coefficient multipliers are respectively interconnected to process input signal samples in the same sequence as like coefficient multipliers process the samples in the FIGURE 2 filter to generate +1 and +Q filtered samples. The output products are applied to a summing circuit 200 which may be similar to the adder tree 70 in FIGURE 2. The FIGURE 3a filter does not require the filtered signal to be processed through an output latch as does the FIGURE 2 circuit to expand the two filtered output samples I„, Q_F over the time period of the four input samples.

In FIGURE 3a the FIR filter includes a "P" register having five cascaded delay stages, P1-P5, into which the positive I and samples are successively loaded. A "M" register including five delay stages M1-M5 is loaded with successive "minus" I and input samples. Each of the delay stages of the P and M registers have output taps for applying delayed signal samples to the respective weighting circuits C1-C9. Certain ones of the weighting circuits are connected directly to the register tap points while others are multiplexed between tap points on the two registers at one half the input sample rate by a multiplexer or switch circuit 2Q6.

FIGURE 3b is a state table showing the states of the particular register stages P1-P5 and M1-M5 for a time TI, and the time T2 which corresponds to the time TI plus one data shift by the registers. The two rows of values C. correspond to the respective weighting circuits connected to the respective register stage in the particular columns for times TI and T2. With reference to FIGURE 3b, FIGURE^" 2 and FIGURE 6, it will be seen that the particular samples of this sequence of nine samples in the P and M registers are applied to like coefficient multipliers as the first sequence of nine samples in the FIGURE 2 filter circuit (as well as all succeeding sets of nine samples). Thus, on a sample by sample basis, the filter functions of the FIGURE 2 and 3a filters are the same.

Loading of the P and M registers is controlled by the signal sampling circuits 201 and 202 respectively, and the register clocking pulses generated by gate 203. The sampling circuits apply all of the positive I and Q signal samples to the P register and all of the negative I and samples to the M register. Because the two registers operate in parallel, they can be operated at 1/2 the input sample rate. However, because two successive samples at the 4fsc rate are alternately applied to each register i.e., +1 and +Q to register P then -I and - to register M, each second sample (the ± samples) must be temporarily latched (204,205) before application to the respective register so that data is not lost.

The timing of the clocking signals applied to sampling circuits 201 and 202 is illustrated in FIGURE 8. A detailed discussion of the operation of the FIGURE 3a type filter may be found in U.S. Patent Application Serial Number 405,173 entitled "Decimating And Demodulating FIR Filters As For TV Chroma Signals" by H. G. Lewis Jr., filed August 4, 1982. In FIGURE 4, the comb filter 20 of FIGURE 1 is shown in further detail. The filtered signal l_, Q_ from the bandpass filter 12 is coupled to the first stage τ_cl of a 454 stage shift register 22, and to an input of an adder 28. The shift register 22 is clocked by an I/

CLOCK-__ signal at a 2f„sc„ rate, producing samples delayed by the scanning time of one horizontal television line at the output of the last stage τ_C454. ' The delayed samples are to be subtractively combined with the undelayed samples in subtractor 24. The delayed samples are arithmetically negated and added to the undelayed samples. The process of arithmetically negating digital samples is the two's complement operation. To perform this operation, the bits of the delayed samples are inverted or one's complemented by inverting circuit 26, which inverts each individual sample bit. The one's complemented delayed signal is then applied to a second input of adder 28, together with a "carry-in" CI equal to '1'. The addition of the CI bit accomplishes a two's complementing of the delayed signal for proper two's complement subtraction. The comb filtered signals at the output of adder 28 are latched into latch L_CF by the I/Q CLOCK signal. Adder 28, like the adders of the bandpass filter, is assumed to exhibit a 40 nanosecond propagation delay. Thus, a comb filtered signal sample ^Iιr_r7 -*-^s Produced at the output of latch L _ delayed by one cycle of the I/Q CLOCK with respect to corresponding signal sample I_1F, as shown by FIGURE 6f. It is seen that filtered sample I-_F is produced at time t_g in FIGURE 6e, whereas the corresponding comb filtered signal sample I_1CF is produced at time t,_Q in FIGURE 6f.

FIGURE 5 illustrates an interpolator suitable for use as interpolator 16 in the embodiment of FIGURE 1. Any of a number of interpolation techniques may be employed to create a 4f rate chrominance signal from the 2 rate chrominance signal produced by the comb filter

20. The interpolator shown is a linear interpolator. This circuit may be replaced by a variety of polynomial interpolators depending on the particular transient response desired of the combed luminance signal. For a discussion of digital interpolation, see R. E. CROCHIERE and L. R. RABINER, "Interpolation and Decimation of Digital Signals - A Tutorial Review", Proc. of IEEE, Vol. 69, No. 3, March 1981, pp. 300-331. The interpolator of FIGURE 5 implements an algorithm of the form

-CI

^■1= 1⁺ V (Q

-Q= V

where I, and Q-, are comb filtered chrominance samples I-_>CF and Q,lC-,„F, and Io and Qo are comb filtered chrominance samples preceding samples I_1CF and Q_1CF- A multiplexer then generates a combed chrominance data stream of the form I,, Q-,, -I-,. " -*/ ^{at a 4}*^_ΞC rate.

In FIGURE 5, the output I_CF, Q_CF of the comb filter latch L_cp (of FIGURE 4) is coupled to two delay stages 80 and 81 which are clocked by the I/Q CLOCK signal, and to an input of an adder 82. The output of the second delay stage 81 is coupled to a second input of adder 82. The output of adder 82 is coupled to the input of an inverting circuit 83, the output of which is coupled to one input of an adder 84, the other input of which receives the value '1'. The output of adder 84 is coupled to the input of a delay stage 86. Stage 86 is also clocked by the I/Q CLOCK signal. The output of delay stage 86 is coupled to inputs of AND gates 96 and 98, and the output of comb filter latch L_,_F is coupled to inputs of AND gates 92 and 94. The second inputs of respective AND gates 92-96 are coupled to receive respective phase-shifted clock signals I₄, Q₄, -I₄ and -Q₄- The outputs of AND gates 92-96 are coupled to inputs of an OR gate 90 to form a multiplexer which produces a combed chrominance signal at a 4f rate. The output of OR gate 90 is coupled to the data input of a D-type flip-flop 100 which has its clock input coupled to receive the 4F clock signal. The interpolator operates to form interpolated values -I and -Q by adding I-- and Q_CF samples to previous I L-„i and Q„ m,mE- samples, respectively, in adder 82. For instance, when the L__F latch is applying sample I-,_CF to one input of adder 82, the preceding sample Q_QCF is held in stage 80 and the preceding sample I_QCF is held in stage

81 and applied to the second input of the adder 82. The next I/Q CLOCK cycle will apply samples Q_1CF and Q_0CF to the adder 82. The adder 82 thus produces consecutive samp ^cles of the form In+1,n,-1-, and Qn+Q"n„-l, . These samp ^cles are of the same bit length as the input signals to the adder, but exclude the least significant adder output bit. That is, if the input signals to the adder are eight bits in length, the adder produces nine output bits, with the upper eight bits taken as the output samples. This effects a division of the sum by two. The bits of the adder samples are then inverted, or one's complemented by the inverting circuit 83. A value '1' is then added to the inverted sum from inverting circuit 83 by adder 84, which effects a two's complementing or arithmetic negation of the samples. Interpolated samples -I and -Q according to the above algorithm are thus produced at the output of the adder 84. These samples are then clocked sequentially into delay stage 86, which produces a sequence of interpolated samples as shown in FIGURE 6g.

The interpolated -I and -Q samples and the comb filter output samples are then multiplexed at a 4f rate by AND gates 92-96 and OR gate 90. Referring to FIGURE 6h, at time t-,, the I . clock applies sample I_1CF o FIGURE 6f through OR gate 90 to the input of D-type flip-flop 100 by way of AND gate 92. At time t_χ2, the 4F_gc clock signal loads this sample into the D-type flip-flop 100, identified by the I, interval in FIGURE 6i. At time t₁₃, sample Q_1CF of FIGURE 6f is applied through OR gate 90 to the input of flip-flop 100 by the Q . clock signal applied to AND gate 94. At time t-₄, this sample is loaded into flip-flop 100 as shown by interval Q, in FIGURE 6i. At time t₁₅, the -I₁ sample held in stage 86 is applied

OMPI

_\ .._. W1PO through OR gate 90 to the input of flip-flop 100 by way of AND gate 96 and clock signal -I₄- At time t-,₆, the -I, sample is loaded into flip-flop 100, and delay stage 86 is clocked to hold the -Q, interpolated sample. At time t₁₇, the -Q-, sample is applied to OR gate 90 by way of AND gate 98 as enabled by the -Q₄ clock signal, and at time t_ιg the -Q sample is loaded into flip-flop 100. Operation continues in this manner so that the 4f rate signal sequence of FIGURE 6i is produced at the output of flip-flop 100.

The 4f chrominance signal sequence is then subtracted from the composite video signal by subtractor 18, which includes an inverting circuit 102 and an adder 104 as shown in FIGURE 5. The bits of the chrominance samples produced by flip-flop 100 are applied to inverting circuit 102 which, together with the carry-in bit CI being 'I^* of adder 104, accomplishes a two's complementing of the chrominance signal for subtraction. A second input of adder 104 is coupled to receive composite video signal samples from delay stage 14, which in this embodiment is a five-stage shift register clocked by the 4F clock signal. The delay stage 14 aligns samples of the composite signal in time with corresponding chrominance signal samples. Referring again to FIGURE 6a, it will be remembered that sample I, of the composite signal was located in center stage τ₄ of the bandpass filter shift register 30 at time ,. The corresponding I, chrominance sample was not produced by flip-flop 100 until time t-₂, as shown in FIGURE 6i. From FIGURE 6a, it is seen that a delay of nine cycles of the 4Fbv_ι clock occurred between time t_, and time t,₂• Thus, delay stage 14 must delay the composite signal by nine cycles of the 4F clock to bring the samples into proper time sequence at the inputs of adder 104. Four cycles of this delay are provided by stages τ₅, τ₆, τ- and τ_Q of the bandpass filter shift register. The remaining five cycles of delay are provided by shift register stages X_g-t₁₃ in FIGURE 5, which receive composite video samples from shift register stage τ_Q in

- U E

OMP FIGURE 2. Therefore, the signal samples are in a proper time sequence at the inputs of adder 104 so that the combed chrominance samples are subtracted from the composite video signal samples, producing a comb filtered luminance signal at the output of adder 104.

Necessary clock signals for the embodiments of FIGURES 1-5 are produced by the clock generating network of FIGURES 7 and 9. The composite analog video signal is applied to a burst gate 110 in FIGURE 7, which gates the color burst signal to a phase-locked loop 112. The phase-locked loop 112 produces a signal F at the color^' subcarrier frequency in phase alignment with the color burst signal. An F 5C signal waveform is shown in FIGURES

8a and 10a. For I, Q sampling a delay equal to 57° of the F signal would be inserted at the output of the phase-locked loop 112.

An inverter 114 is coupled to receive the F signal and produces a complementary signal F^~ at its output- The F 9G signal is also applied to a phase detector 121 of a second phase-locked loop 120. The phase detector 121 produces a control signal to control the phase and frequency of an oscillator 122, which produces a 4F signal at four times the color subcarrier frequency, as shown in FIGURE 8b. The 4F signal is applied to a divide-by-two circuit 124 and to an inverter 123, the latter producing a complementary 4F 9G signal. The divide-by-two circuit 124 produces a 2F 9G signal at twice the subcarrier frequency, as shown in FIGURE 8c. The 2F 9G signal is inverted by an inverter 126 to produce a complementary 2F signal. The 2F signal is also applied to the input of a second divide-by-two circuit 125, which produces an f 9G rate signal for the phase detector 121. Thus, the Fsc, 4Fsc and 2Fsc signals and their complements are maintained in substantial phase synchronism.

The Fsc, 2Fs-c and 4Fsc sig'nals are applied to inputs of an AND gate 130 to produce pulses of an I-, signal when all of the input signals are high, as shown in FIGURES 8d and 10c . Similarly, the F9G , 2F_ _ and 4F9G signals are applied to inputs of an AND gate 132, which produces a Q_c signal as shown in FIGURES 8e and lOd. The 4F SG and 2FSC signals are applied to inputs of an AND gate

134, to produce an I/Q CLOCK signal, as shown in FIGURES 8f and lOg. The F Ow, 2F C and ?FSG signals are applied to inputs of an AND gate 136 to produce clock signal -I_c which is illustrated in FIGURES 8g ^ and lOe. The Fs.c, 2F_sc,_, and 4F - signals are applied to inputs of an AND gate 138 to produce clock signal -Q_c as shown in FIGURES 8h and lOf. The I/Q CLOCK signal of FIGURE 8f is seen to be in the proper phase alignment with the 4F 9G signal for operation of the bandpass filter, comb filter and interpolator of FIGURES 2, 4 and 5, and the 1^, _c, -I_c and -Q_c clock signals are seen to be in the quadrature phase relationship required by the multiplexer of the interpolator of FIGURE 5. The clock signals of FIGURE 8 are also in the same timing relationships as the signal waveforms of FIGURE 6. By reason of the odd multiple frequency relationship between the line scanning (horizontal synchronizing) frequency and the color subcarrier (f_sc) frequency, samples on one video line of information are of opposite phase with respect to vertically aligned samples on succeeding and preceding lines. That is, if the first sample on one line is a +1 sample, the first sample on the next line is a -I sample. This relationship, which underlies the principle of comb filtering, requires that the phases of the I S_m_τLl and Q(__,_r___ clock signals be changed from line-to-line so that +1 and +Q samples will be selected on one line and combined with vertically aligned -I and -Q samples on the next line. In addition, the interpolator must be operated so that, on the first line, -I and -Q samples are interpolated to fill in between the +1 and +Q samples, and that +1 and +Q samples are interpolated to fill in between the -I and -Q samples on the following line. Clock circuitry which performs the necessary phase shifts of the clock signals is shown in FIGURE 9. In FIGURE 9, a synchronizing signal source 140 produces a horizontal blanking signal, which is applied to the clock 5 input of a D-type flip-flop 150. The data input of flip-flop 150 is coupled to a source of logical one level voltage (+). The Q output of flip-flop 150 is coupled to the data input of a D-type flip-flop 152. The Q output of flip-flop 152 is coupled to the clock input of a D-type

10 flip-flop 154, and the Q output of flip-flop 152 designated SKIP is coupled to the reset input R of flip-flop 150, and input of an AND gate 180, and to the input of the first stage 171 of a SKIP signal delay line 170. The Q output of flip-flop 154 is coupled to an input

15 of an exclusive-OR gate 160, and to the data input of a D-type flip-flop 178. The Q output of flip-flop 154 is coupled to the data input of flip-flop 154. The second input of exclusive-OR gate 160 is coupled to receive the F__r clock signal, and produces a signal LINE at its

20 output. The LINE signal is applied to the clock input of flip-flop 152.

The SKIP signal delay line 170 includes five serially coupled delayed stages 171-175, which are clocked by pulses of the I-, and -I_c clock signals applied by an OR

^'25 gate 177. The output of the third delay stage 173 designated FSKIP is coupled to one input of an AND gate

190, which receives the I/Q CLOCK signal at its other input. The AND gate 190 produces an output signal identified as I/Q CLOCK-j_. The output of the last delay

30 stage 175 is coupled to the clock input of flip-flop 178 by an inverter 176. The Q output of flip-flop 178 is coupled to one input of an exclusive-OR gate 164, which is also coupled to receive the F_sc clock signal. The exclusive-OR gate 164 produces a signal LINED, which is

35 also produced in inverted form by an inverter 166.

The AND gate 180 is coupled to receive the I/Q CLOCK signal at its second input, and produces an output signal identified as I/Q CLOCKg

C PI

^[ Three AND gates 182, 184 and 186 each have an input to the output of OR gate 177. The LINE signal is applied to a second input of AND gate 182, which produces the I_CL clock signal for the L_χ latches of the bandpass 5 filter as its output. The LINED signal is applied to the second input of AND gate 184, which produces the I₄ signal for the interpolator at its output. The LINED signal is applied to the second input of AND gate 186, which produces the -I₄ clock signal for the interpolator at its

10 output.

Three AND gates 192, 194 and 196 each have an input coupled to the output of an OR gate 198. The OR gate 198 is coupled to receive the Q_c clock signal at one input and the -Q_, clock signal at a second input. The

15 LINE signal is applied to a second input of AND gate 192 to produce the Q_CL clock signal for the L_ latches of the bandpass filter. The LINED signal is applied to the second input of AND gate 194, which produces the Q₄ signal for the interpolator at its output. The LΪNED signal is_

20 applied to the second input of AND gate 196 to produce the -Q₄ signal for the interpolator. The output of OR gate 198 is also coupled by way of an inverter 156 to the reset input R of flip-flop 152.

The operation of the clock signal circuitry of

25 FIGURE 9 may be understood by first assuming that flip-flops 150, 152 and 154 are all in a reset condition. With flip-flop 154 being reset, the LINE signal at the output of exclusive-OR gate 160 will be in phase with the F_sc signal, as seen by comparing pulse 200 of FIGURE 10b

30. with pulse 202 of FIGURE 10k. The pulses of the LINE signal of FIGURE 10k gate pulses 204 and 206 of the I- and Q_c signals (FIGURES 10c and lOd) to the L, and L_Q registers 40 of the bandpass filter by way of AND gates 182 and 192 to selectively load the latch registers at

35 these times. This timing will load the latch register 40 at the times of occurrence of +1 and +Q signal samples of the incoming sample sequence, as seen by comparing FIGURES 10b and 10c with the sample sequence shown in FIGURE 10a. When the video signal is sampled by the A/D converter 10 at a 4f_s_ rate, each horizontal line of an NTSC signal will be represented by 910 samples. The LINE signal will thus load the latch register 40 during the times of occurrence of the first and second samples of each group of four samples, including the 905th and 906th samples shown in FIGURE 10a.

The 910 samples of one line of the NTSC signal are aligned in time with 227^*% cycles of the F_sc signal of FIGURE 10a. In order to operate the comb filter 20 substantially at a 2f_sc rate and to comb filter by combining vertically aligned samples from line-to-line, the extra half line of samples, shown as 909 and 910 in FIGURE 10a, are deleted from the comb filtering process. At this time, the phase of the LINE signal is reversed to select samples of opposite polarity of the new line for the comb filter. This results in a 454 stage comb filter 22 as shown in FIGURE 4.

The phase change is initiated by the leading edge of the horizontal blanking signal shown in FIGURE lOh, which sets flip-flop 150. The leading edge of the horizontal blanking signal is illustratively shown in FIGURE 10 to occur during sample 905 of FIGURE 10a where FIGURE 10a represents the sample sequence at the stage τ₄ (center tap) of the bandpass filter. Flip-flop 150 applies a high signal to the data input of flip-flop 152, which is then set by the next leading edge 210 of the LINE signal, as shown in FIGURE 10k. The Q output of flip-flop 152 produces a SKIP pulse, shown in FIGURE lOi, which clocks flip-flop 154 to a set condition. The Q output of flip-flop 154, labelled CHANGE, changes state as shown in FIGURE 10j , which changes the phase of the LINE signal, as shown in FIGURE 10k. The LINE signal is now low during the occurrence of I_ and Q_c pulses 214 and 216, preventing loading of the latches 40 during samples 909 and 910 of FIGURE 10a. In addition, the SKIP pulse at the Q output of flip-flop 152 blocks the passage of I/Q CLOCK pulse 212 of FIGURE lOg through AND gate 180 so that the multiplexer 50 of FIGURE 2 will not change state during samples 909 and 910. The SKIP pulse also resets flip-flop 150, and is loaded into the first stage 171 of the SKIP signal delay line 170 by I_c clock pulse 214. Flip-flop 152 is reset by Q clock pulse 216, thus terminating the SKIP and SKΪf pulses.

The new phase of the LINE signal will now load the latch registers 40 during the occurrence of -I and -Q samples. LINE pulse 224, for instance, will gate ~I„ and -Q_c pulses 220 and 222 of FIGURES 9e and 9f to the latch registers by way of AND gates 182 and 192. Pulses 202 and 222 are seen to be coincident with -I sample 1 and -Q sample 2 of the new line of samples. These samples are vertically aligned with +1 sample 1 and +Q sample 2 of the previous horizontal line.

It is remembered from FIGURES 6d and 6e that two cycles of the I/Q CLOCK signal are required for the multiplexed signals to propagate through the bandpass filter of FIGURE 2 to the filter output latch L_∑. Since no samples were loaded into the latches 40 of the filter during sample times 909 and 910, it is necessary to prevent operation of the L_∑ latch at the time at which samples from times 909 and 910 would be expected to appear at the input of latch L^.. This is accomplished by the SKIP signal delay line, which clocks the SKIP signal to the output of stage 173 at this time. This signal, termed FSKIP in FIGURE 9, prevents the passage of an I/Q CLOCK pulse through AND gate 190 so that the L_ latch, which is clocked by signal I/Q CLOCK_F, is not latched when filtered sample 909 would be expected to appear at the output of the bandpass filter. The I/Q CLOCK-, signal also clocks the comb filter delay line 20 of FIGURE 4, which similarly skips a clock cycle at this time.

After two more I/Q CLOCK cycles, the skipped sample interval will have reached the interpolator 16 of FIGURE 5. This requires a change of the phase of the I₄, Q₄, -I₄ and -Q. clocks of the interpolator. The SKIP pulse at this time is clocked to the output of the last stage 175 of the SKIP signal delay line 170. The SKIP pulse is inverted and clocks flip-flop 178 to load the CHANGE signal level into flip-flop 178. The CHANGE signal level then reverses the phase of the LINED and LINED signals at the outputs of exclusive-OR gate 164 and inverter 166, respectively. The LINED and LINED signals control AND gates 184, 194, 186 and 196, which generate the I₄, Q₄, -I₄, and -Q₄ clock signals. During the previous line, the I₄ signal and the Q₄ signal were in phase with I_ and Q_ clock signals to gate comb filtered +1 and +Q signals through the interpolator gate 90. The -I₄ and -Q₄ signals were in phase with the -I_c and -Q_, clock signals to gate interpolated -I and -Q signals through gate 90. With the reversal of phase of the LINED and LINED signals, the I₄ and Q₄ clock signals will now be in phase with the -I_c and -Q_c clock signals to gate comb filtered -I and -Q samples through the interpolator, and the -I₄ and -Q₄ clock signals are in phase with the I_ and Q_c clock signals to gate interpolated +1 and +Q samples through gate 90. The interpolator will therefore gate properly phased samples to subtractor 18 for comb filtering of the luminance signal.

The right-hand waveform of FIGURE 10 show that similar clock signal phase reversals occur at the end of the new horizontal line, during which time the CHANGE signal of FIGURE lOj is changed to its reset condition. Analysis of the illustrated embodiment of the present invention has shown that erroneous interpolated samples are produced at the time of clock phase changes from one line to the next. This is because samples of one line are combined with improper samples of the adjacent line by the interpolator. Specifically, the last two interpolated samples of the ending line and the first two interpolated samples of the new line will be in error, which span a time interval of six 4F clock cycles, beginning at the changing of the CHANGE signal of FIGURE lOj . By comparing FIGURE lOh with FIGURE lOj and FIGURE 10a, it is seen that approximately four sample intervals (i.e., four 4F_SC clock cycles) occur between the leading edge of horizontal blanking and the change of the CHANGE signal. Thus, a total of ten cycles of the 4F_SC clock pass between the leading edge of the horizontal blanking pulse and the time at which a steady stream of correct samples resumes. In the NTSC system, this is a time interval of approximately 700 nanoseconds. Since the "front porch" of the horizontal blanking interval which precedes the horizontal sync pulse has a minimum duration of approximately 1.27 microseconds, it is seen that the erroneous samples will have passed through the system before the time of occurrence of the horizontal sync pulse, which is the next video information signal which must be accurately reproduced. Moreover, by effecting the clock phase reversal during the beginning of the horizontal blanking interval, the changeover occurs during a time at which no chrominance information is being transmitted. Thus, the outputs of the bandpass filter and the comb filter will contain substantially no chrominance signal information and essentially zero level signals will be produced by the comb filter. The interpolation process will then be performed entirely with zero level signals and will cause no erroneous results when the zero level signals are introduced into the luminance channel for luminance signal combing.

An important aspect of the present invention is the ability to comb filter the chrominance signals at a sampling frequency which is less than the sampling frequency of the A/D converter. In the embodiment of

FIGURE 4, the comb filter sampling frequency is 2f G, or approximately 7.16 MHz in the NTSC system. Since the NTSC chrominance signal bandwidth extends from approximately 2.1 MHz to 4.2 MHz, it would ordinarily be anticipated that sampling at 7.16 MHz would cause aliasing around the 3.58 MHz color subcarrier frequency. Note, however, that each of the processed samples was originally sampled at a rate that satisfies the Nyquist criterion and from line to line corresponding samples have a precise phase relationship. If the combing function is evaluated in the time domain, it is recognized that the samples of one line are subtracted from corresponding samples of an adjacent line sampled at precisely the same point in their respective lines.

Like components of luminance signal occurring in the band passed samples of adjacent lines are removed by the subtractive process. Further, with respect to the chrominance component of the band passed samples, because the subsampling clocks are locked to the chrominance carrier, the aliasing concept in general does not apply. Sampling the modulated carrier, at the carrier frequency, at the same phase position each cycle satisfies the Nyquist criterion with respect to the chrominance signal and aliasing components are not generated.

In FIGURE 11 another embodiment of a signal separation system according to the principles of the present invention is shown. Elements shown and discussed in conjunction with the embodiment of FIGURE 1 are shown in the embodiment of FIGURE 11, bear the same reference numerals, and perform the same function.

The 4f 9G rate digital samples produced by the

A/D converter 10 are applied to inputs of a delay stage 317 and a digital bandpass filter 312. The passband of the bandpass filter 312 for NTSC video signals, includes the chrominance frequencies which are distributed about the 3.58 MHz color subcarrier frequency. The delay stages 317 provides an equalizing delay which matches the delays encountered by signals between the input of the bandpass filter 312 and the output of a bandpass filter 332. Bandpass filtered signals at a 4f rate sequence are produced at the output of the bandpass filter 312. This sequence is subsampled by a CI sampler 314 and a C2 sampler 316, which pass samples at two different phases of an f 9G rate signal, respectively. The samplers thus subsample the 4f 9G rate sequence at a color

OMPI ^' subcarrier ( f G ) rate in response to the differently phased sampling signals Φ-_jF- and Φ₂ ^F _SC-

Output signals from the samplers 314 and 316 are applied to a CI comb filter 320 and a C2 comb filter 322, respectively. The comb filters 320 and 322 comb filter the applied signals, producing comb-filtered samples at an f SC, rate. The comb filters are respectively clocked by the same Φ m-l.m,F and Φ,__F9G sampling signals as the samplers.

The output samples of the comb filters are then combined in an interleaved sequence of samples at a 2f 9G rate by sequencer 324. This sample sequence is then applied to an interpolator 330, which produces a sequence of samples at the original rate of 4f 9G. This sample sequence is filtered by a bandpass filter 332 so as to pass only samples in a given passband of comb-filtered samples. The bandpass-filtered samples are then subtracted from the delayed composite video signal sample sequence passed by the delay stage 317, resulting in the production of a second sequence of comb-filtered samples. In the embodiment of FIGURE 11, comb-filtered chrominance signals in the chrominance passband are produced at the output of the bandpass filter 332. These signals are then subtracted from the composite video signal in subtractor 18 to produce comb filtered luminance signals. The arrangement of FIGURE 11 is advantageous in that the differently phased comb filters 320 and 322 are operated at the subcarrier frequency. For NTSC signal samples, each comb filter will include a one-H delay line clocked at the subcarrier frequency of approximately 3.58 MHz, resulting in a one-H delay line length of 227 stages per filter. For example, if the samples applied to the comb filters are eight bits in length, each one-H delay line will require only 1816 storage locations, or a total of 3632 storage locations for both comb filters. This compares favorably with the 7280 storage locations required for a comb filter operating on eight-bit samples at a 4f G rate. Design of the system is simplified in that the two bandpass filters 312 and 332 can be FIR filters of the same order using the same weighting coefficients. In addition, the bandpass filter 332 may be arranged to provide the interpolation function, as will be subsequently described. A digital bandpass filter suitable for use as bandpass filter 312 in the embodiment of FIGURE 11 is shown in detail in FIGURES 12a and 12b, which interconnect at the broken connections at the sides of the FIGURES. In FIGURES 12a, 12b, and 14, thin lines depict data paths for multibit samples and broad lines depict clock signal paths.

The bandpass filter of FIGURES 12a and 12b comprises a 31-stage output tapped FIR filter, including a shift register 340 of thirty delay elements ^τ'ι"^t'_3θ' thirty-one coefficient weighting circuits 350, and an adder tree 360. Samples of the video signal produced by the A/D converter are applied to the input of the first stage τ' of the shift register 340, and are shifted through the register by a 4F 9G clock signal. Samples are tapped from the input of the first stage and the output of each stage of the shift register and are applied to the coefficient weighting circuits 350. Precise coefficient values are shown in FIGURES 12a and 12b (-.0277, +.0485, +.0113, etc.), which may be implemented using coefficient multiplier circuits which multiply each tapped signal by the appropriate coefficient value. If desired, the coefficient weighting circuits can also be implemented using a shift-and-add technique by rounding off the coefficient values to the nearest multiple of an inverse the output of stage τ ', has a value of +.0485. The nearest multiple of an inverse power of two is 3/64, which is equal to .046875. Weighting by this value is achieved by shifting the tapped signal to the right by six places, which weights the tapped signal by 1/64, and by shifting the tapped signal to the right by five places, which weights the signal by 1/32. The two shifted signals are then added to produce a value equal to 3/64 times the tapped signal. Circuitry for performing the shift-and-add

.WIPO function is shown in United States patent application number 363,827, entitled "Digital Filter Circuits" by Lauren A. Christopher, filed March 31, 1982.

The tapped and weighted signals are applied to the adder tree 360, in which they are combined by the adders to produce a filtered signal at the output of the final adder 361. It is essential that the tapped and weighted signals all arrive at the output of the final adder 361 at the same time. To insure that this will happen, each signal from a coefficient weighting circuit is routed to the output through six adders, or delays equivalent to six adders. It may be seen that tapped signals from shift register stages

encounter fewer than six adders on their passage from the coefficient weighting circuits to the output of adder 361. To equalize the propagation delays of these signal paths with respect to the remaining signal paths, delays 362-364 equivalent to the propagation time of one adder tree adder and a delay 365 equivalent to the propagation time of three adders are inserted into these shorter paths. Thus, the adder tree will delay each tapped and weighted signal by the propagation time of six adders between the coefficient weighting circuits and the output of adder 361. The filtered signals at the output of adder 361 are selectively sampled by latches L, and L_Q in response to clock signals I_F_ and Q_FC- The I_FC and Q_FC clock signals each have a frequency of f G (3.58 MHz in the NTSC system) to sample one of every four filtered signal samples. The I_FC and Q_FC clock signals also sample the power of two. For instance, the weighting coefficient at filtered signals at respectively different phases of the color subcarrier signal and in this embodiment exhibit a 90° phase relationship with respect to each other. The signal samples in the L_χ and L_Q latches are then applied to two comb filters 320 and 322. Each comb filter includes a 227 stage delay line 370 and 380, which delays the applied signals by one horizontal line

O PI interval. The delay lines 370 and 380 are clocked by the same I_Fr and _FC clock signals that are used by the L, and L_Q latches. The bits of the delayed samples are inverted, or one's complemented by inverting circuits 372 and 382 and applied to inputs of respective adders 374 and 384. Respective undelayed signals are also applied to the adders 374 and 384. A least significant "carry-in" bit CI equal to 'I' is also applied to each adder which, together with the one's complement of the delayed samples, accomplishes a two's complementing or arithmetic negating of the delayed signals for subtraction. The adders 374 and 384 will thereby effectively subtract the delayed samples from the undelayed samples to produce comb filtered signals at their outputs. These comb filtered signals are latched into respective latches L_1CF and ~_CF by differently phased f rate clock signals I_CFC and

Q_CFC. Thus, sequences I -. and Q_CF of comb filtered signal samples are produced at the outputs of latches L_ICF and

L_ry._r,-__r_-. _' These signal samples are available for subsequent processing in a chrominance signal processor to produce demodulated color mixture signals.

In FIGURE 14, the comb filtered chrominance signal sequences I__F and Q_CF are used to produce a comb filtered luminance signal. The outputs of the _ICF and ^L _F latches are coupled to inputs of respective AND gates 390 and 392. The AND gates 390 and 392 are also coupled to receive complementary clock signals 2F^~ _ and ^2F _SCD' respectively. A third clock signal, LINE,_™,, is applied to third inputs of AND gates 390 and 392. The outputs of AND gates 390 and 392 are coupled to the inputs of an OR gate 394, the output of which is coupled to the input of a bandpass filter.

The purpose of the bandpass filter of FIGURE 14 is to create a sequence of comb filtered chrominance signals in the chrominance passband at a 4f clock rate. The bandpass filter of FIGURE 14 is constructed in the same manner as the input bandpass filter of FIGURE 12, being of the same filter order with the same weighting

OMPI - / V/IFO coefficient values. The bandpass filter of FIGURE 14 includes a thirty stage shift register 440, thirty-one coefficient weighting circuits 450, and an adder tree 460 having a final output adder 461. For ease of illustration, only a portion of these circuit elements are shown in FIGURE 14.

The shift register 440 of the bandpass filter is clocked at a 4f G rate by the 4F clock signal. Bandpass filtered output signals from the final adder 461 are latched into latch

by the 4FSC clock signal. The 4f9G rate sample sequence produced by latch L_4χC comprises comb filtered chrominance information in the chrominance passband, which is also available for subsequent demodulation and processing. The bits of each sample in the 4f 9G rate sequence are one's complemented, or inverted, by an inverting circuit 396. The one's complemented samples are applied to an adder 400 together with a carry-in bit CI equal to 'l¹ to form two's complemented or arithmetically negated chrominance signals for subtraction from the composite video signal.

Composite video signal samples are applied to a second input of the adder 400 by way of a fifteen stage shift register delay line 398 having the input to the first stage τ"₃₁ coupled to the output of the shift register 340 of the input bandpass filter. The shift register 398 is clocked by the 4F clock signal and acts to present composite video signal samples to the adder 400 in time coincidence with corresponding comb filtered chrominance signal samples. The adder 400 thus produces a comb filtered luminance signal which has been combed over the chrominance passband. The comb filtered luminance signal is available at the output of adder 400 for subsequent luminance signal processing.

Clock signals for the embodiment of FIGURES 12 and 14 are produced by the clock generating circuitry of FIGURE^'S 7 and 13. FIGURE 7 (described above) supplies signals to the generator circuitry of FIGURE 13. The

OMPI clock generator of FIGURE 13 is similar to that of FIGURE 9.

As discussed in conjunction with FIGURES 7 and 9, the various sampling clocks must be appropriately phased for proper operation in an N.T.S.C. system. Clock circuitry which performs the necessary phase shifts of the clock signals from one line to another is shown in FIGURE 13. In FIGURE 13 a synchronizing signal source 140' produces a horizontal blanking signal, which is applied to the clock input of a D-type flip-flop 150' . The data input of flip-flop 150' is coupled to a source of logical one level voltage (+). The Q output of flip-flop 150' is coupled to the data input of a D-type flip-flop 152'. The Q output of flip-flop 152' is coupled to the clock input of a D-type flip-flop 154' and the Q output of flip-flop 152' is coupled to the reset input R of flip-flop 150'. The Q output of flip-flop 154' is coupled to an input of an exclusive-OR gate 160'. The Q output of flip-flop 154' is coupled to the data input of flip-flop 154' . The second input of exclusive-OR gate 160' is coupled to receive the F 9G clock signal, and produces a signal LINE at its output. The LINE signal at the output of exclusive-OR gate 160' is applied to the input stage of a ten-stage shift register 212. The shift register 212 is clocked by a signal 2^~F^~ _D, which is a phase-shifted replica of the 2F 9G signal of FIGURE 8c

A signal LINE_p is produced at the output of the shift register 212 and is applied to the input of a flip-flop 214. The output of flip-flop 214, which produces a signal termed LINE_CF, is applied to the input of a flip-flop 216, the output of which is coupled to the input of a further flip-flop 218. Flip-flops 216 and 218 are clocked by the 2F^~ _D clock signal produced at the Q output of a D-type flip-flop 220. The flip-flop 220 receives the 2F signal at its data input and the 4F^~ clock signal at its clock input.

The I_c and -I_c signals are applied to the inputs of an OR gate 177. The output of OR gate 177' is coupled to inputs of AND gates 232 and 234. The LINE_p signal is applied to the second input of AND gate 232, which produces a signal I_FC at its output. The LINE_CF signal is applied to the second input of AND gate 234 to produce a signal Q_CFC at the output of that gate.

The Q_c and -Q_c signals are applied to the inputs of an OR gate 198' . The output of OR gate 198' is coupled to an input of an AND gate 242. The signal produced at the output of OR gate 240 is also used to clock flip-flop 214. The LINE-, signal is applied to a second input of AND gate 242, which produces a signal identified as Q_F_, and I__F_ at its output.

When the A/D converter of FIGURE 11 samples the composite video signal at a 4fbv rate, each line of the NTSC signal will comprise 910 samples. FIGURE 15a illustrates the last seven samples of one line of information, which line begins with a +1 sample (not shown) and ends with a +Q sample 910. The succeeding line begins with a -I sample 1, and ends with a -Q sample (not shown) . The reversal of the clock phasing for proper comb filtering is done during the last two samples of each line, and is initiated by the leading edge of the horizontal blanking signal shown in FIGURE 15g. The sample sequence of FIGURE 15a represent the samples at the input to the bandpass filter shift register 40 of FIGURE 12.

Initially it will be assumed that flip-flops 150', 152' and 154' are all reset. Under these conditions, the LINE signal at the output of exclusive-OR gate 160' will be in phase with the F signal, as seen by comparing pulse 500 of FIGURE 15b with pulse 502 of the LINE signal of FIGURE 15h. The leading edge of the horizontal blanking signal of FIGURE 15g, illustratively shown to occur during codeword 905 of the line that is ending, sets flip-flop 150' at time t₁ in FIGURE 15g. The next rising edge of the LINE signal occurs at time t„ in FIGURE 15h, which will set flip-flop 152' at that time.

OMFI The signal at the Q output of flip-flop 152' is shown in FIGURE 15j .

The setting of flip-flop 152' in turn sets flip-flop 154*, causing the signal at its Q output to go high. This results in a reversal of the phase of the LINE signal following time t₂, as shown in FIGURE 15h. The pulses of the LINE signal are now coincident with pulses of the -I_c and -Q_c clock signals of FIGURES 15e and 15f, whereas previously they were coincident with the I_c and Q_ pulses of FIGURES 15c and 15d. Flip-flop 150' is reset at time t₂ by, the low-going signal at the Q output of flip-flop 152', and flip-flop 152' is reset at time t₃ by a low-going signal (shown in FIGURE 15i) produced by a NAND gate 244, which is coupled to receive the 4F 9G and 2F^~ clock signals.

At the time that the LINE signal changes phase, no immediate change in system operation occurs, since the final samples of the ending line are just entering the shift register 340 of the input bandpass filter of FIGURE 12. Fifteen cycles of the 4F clock are required for the samples to reach the impulse response center of the bandpass filter at the output of shift register stage τ-,5⁴ Tapped signals must then propagate through the coefficient weighting circuits 350 and the adder tree 360 before the filtered signal is available to be latched into the L- and L₀ latches. For purposes of illustration, it is assumed that the coefficient weighting circuits 350 each exhibit a propagation time of 66 nanoseconds, and that each adder in the adder tree exhibits a propagation time of 40 nanoseconds. Since each tapped signal in the bandpass filter must pass through one coefficient weighting circuit and six adder delays, the tapped signals are delayed by approximately 306 nanoseconds between the outputs of the shift register stages 340 and the output of the final adder 361. Thus, filtered samples appear at the output of adder 361 in the sequence shown in FIGURE 15k with respect to the bandpass filter input sample sequence shown in FIGURE 15a. The sample sequence of FIGURE 15 is delayed with respect to the sample sequence of FIGURE 15a by more than 19 cycles of the 4F clock.

At the same time as the samples at the end of the line are propagating through the bandpass filter, the LINE signal is being shifted through the ten stage shift register 212 of FIGURE 13 by the 2F^~ _D clock signal. A delayed LINE signal, termed LINE_p, is produced at the output of shift register 212 and is shown in FIGURES 15m and 16c. The input sample sequence of FIGURE 15a is redrawn in FIGURE 16a the filtered sample sequence of FIGURE 15k is shown again in FIGURE 16d, and the 2F _D clock signal is shown in FIGURE 16b.

During the occurrence of positive filtered I and

Q samples of the ending line, such as +I-, sample 905 and +Q_p sample 906 of FIGURE 16d, the high LINE_p signal of

FIGURE 16c gates sampling pulses through AND gates 232 and 242 of FIGURE 13. These pulses, identified as I_FC and Q_FC in FIGURES 16e and 16f, load the +I_, and +Q_F samples into latches L_τ and L_Q, respectively. During the occurrence of filtered negative I and Q samples of the new video line, such as -I_p sample 1 and -Q_F sample 2 of FIGURE 16d, the phase of the LINE_F signal has changed so that the LINE-, signal is high during these negative signal samples. During the new line, LINE„r pulse 510 will gate sampling pulses 512 and 514 of the I_. and Q_p clock signals to latches L, and L₀ to clock the negative sample into the latches. The latched samples thus alternate between positive samples on one line and negative samples on the following line. The I_F and Q_FC signals also clock the comb filter delay line registers 370 and 380 of FIGURE 12b. The samples at the outputs of the respective latches and delay line registers are of opposite polarity, and are subtractively combined in adders 374 and 384. The adders 374 and 384 are also assumed to exhibit propagation delays of 40 nanoseconds, and produce sequences of comb filtered samples shown in respective FIGURES 16g and 16h, delayed with respect to clock signals I„_ and Qt-,v__. by the 40

O PI nanosecond adder delay. The further delayed LINE signal produced by flip-flop 214 of FIGURE 13, identified as the LINE_CF signal shown in FIGURE 16i, gates pulse Q_CFC of FIGURE 16k through AND gate 242. The I_CFC and Q_CFC pulses of FIGURES 16j and 16k load the comb filtered samples of FIGURES 16g and 16h into latches L_ICF and L_0CF of FIGURE 12b, producing the comb filtered sequences of FIGURES 16m and 16n at the outputs of these latches.

The comb filtered samples are now interpolated to a 4f 9G rate sample sequence by the arrangement of

FIGURE 14. The INE_cp signal of FIGURE 16i is delayed by flip-flops 216 and 218 of FIGURE 13 to produce a signal LINE_INT of FIGURE 16p. The LINE,_NT signal is applied to AND gates 390 and 392 of FIGURE 6 to allow passage of positive samples through the gates during one line of video information, and negative samples through the gates during the following line. When the F_SCD signal of FIGURE 16b is low (and the 2F^~ _ signal is high), a comb filtered I_CF sample is passed by AND gate 390 and appears at the output of OR gate 394, as shown by the +1 and -I sample intervals of FIGURE 16q. When the 2F_SCD signal is high, a comb filtered Q_CF sample is passed by AND gate 392 and appears at the output of OR gate 394, as shown by the +Q and -Q sample intervals of FIGURE 16q. And when the L NE_TNT signal is low, a zero value signal level is produced at the output of OR gate 394, as indicated by the zero value intervals of FIGURE 16q.

The output sequence of OR gate 394 is shifted into the shift register 440 of the output bandpass filter of FIGURE 14 by the 4F 9G clock signal. The sequence appearing at the output of the first stage t" of register 440 is shown in FIGURE 16r. The output bandpass filter delays the applied signal in the same manner as the input bandpass filter of FIGURE 12 by the time required for the samples to reach the impulse response center (center tap) of the filter and then to propagate through the coefficient weighting circuits 450 and the adder tree 460. By virtue of the polarity and magnitude of the coefficient

OMPI values, the output bandpass filter interpolates -I and -Q values in the zero signal intervals of the ending line, and +1 and +Q values in the zero signal intervals of the new line. The sequence at the output of the final adder 461 is latched into the output latch L_4xC ^bY "the ^4F _Sc clock signal, producing a 4f 9G rate output sequence of samples shown in FIGURES 16s and 17a. This sequence comprises comb filtered chrominance signals in the chrominance passband defined by the bandpass filter response.

The comb filtered chrominance signal is then subtractively combined with the composite video signal to produce a luminance signal which is devoid of chrominance components. The composite video signal sample sequence at the input to the system is redrawn as FIGURE 17b, and is seen to lead the comb filtered chrominance sequence by 43 cycles of the 4F clock signal. For proper combing, the composite video sequence must be brought into time coincidence with corresponding samples of the chrominance sequence. This is done by delaying the input composite video sequence by 43 cycles of the 4F 9G clock. A thirty clock cycle delay is provided by the shift register 340 of the input bandpass filter, which produces a composite video signal sequence at the output of stage τ"_3Q delayed by thirty clock cycles with respect to the input sequence. The remaining thirteen cycles of delay are provided by shift register delay line 317 of FIGURE 14, which is also clocked by the 4F clock. Thus, the sample sequences at the inputs of the adder 400 are in the proper time alignment for the production of a comb filtered luminance signal.

It is seen in FIGURE 16q that two zero value samples are interspaced between each pair of I and Q samples that are applied to the output bandpass filter, except at the end of the line when the LINE signal phase is changed. At that point there are four consecutive zero value samples. It might be expected that this change of the sample sequence would cause an error in the

- UR

O reconstituted chrominance signal, and hence an error in the comb filtered luminance signal. However, since the phase change is initiated by the horizontal blanking signal, the phase change occurs during the "front porch" interval of the horizontal blanking interval. At this point in the video signal there is no chrominance signal content, and hence the true chrominance signal values are zero. Thus, the insertion of zero values into the sample sequence at this point will cause no errors in the resulting comb filtered signals.

A second embodiment of the present invention is illustrated in FIGURE 18. Elements shown and discussed in conjunction with the embodiment of FIGURE 11 are shown in the embodiment of FIGURE 18 and bear the same reference numerals. Whereas a comb filtered luminance signal was produced in FIGURE 11 by first comb filtering the chrominance signal, then subtracting the comb filtered chrominance signal from the composite signal, the comb filtered luminance signal in FIGURE 18 is produced by comb filtering a high frequency portion of the composite video signal to produce a comb filtered high frequency luminance signal portion, then adding this signal portion to a low pass filtered portion of the composite signal.

In FIGURE 18, the composite video signal is first filtered by a bandpass filter 312, leaving a passband occupying frequencies about the color subcarrier frequency. The bandpass filtered signal is then subsampled by CI and C2 samplers 314 and 316, in the same manner as FIGURE 11. Signal components produced by the CI sampler 314 are applied to a first comb filter 520, and signal components produced by the C2 samples 316 are applied to a second comb filter 522. Both subsampled signal sequences are produced at an f rate, but are in respectively different phase relationships with respect to the subcarrier signal phase, since they are produced by sampling signals φ, 1Fsc and φ2₀F„sc.

The signal sequence produced by the CI sampler 314 is applied directly to inputs of an additive combining

-^!&U E

OMPI circuit 578 and a subtractive combining circuit 576, and is applied by way of a one-H delay line 570 to second inputs of the additive and subtractive combining circuits. The one-H delay line 570 is constructed in the same manner as the shift register delay line 370 of FIGURE 12.

The signal sequence produced by the C2 sampler 16 is applied to inputs of a one-H delay line 580, an additive combining circuit 588, and a subtractive combining circuit 586. The output of the one-H delay line 580 is coupled to second inputs of the additive and subtractive combining circuits 588 and 586.

Circuits 576 and 586 subtractively combine delayed and undelayed samples to produce comb filtered chrominance signal sequence at two phases φ. and Φ₂ of the color subcarrier (F U) signal, in the same manner as adders 374 and 384 of FIGURE 12b. Circuits 578 and 588, by additively combining the delayed and undelayed sample sequences, produce samples in which the chrominance components have been cancelled, leaving comb filtered luminance signal components of two different subsampling phases of the 4fsc A/D converter clock. These comb filtered luminance signal samples occupy the passband established by the bandpass filter 312 and are therefore high frequency luminance signal components. The differently phased luminance samples produced by additive combining circuits 378 and 388 are interleaved in a single sample sequence by a sequencer 324, which sample sequence is at a 2f SC rate. The 2f9G sample sequence is interpolated by interpolator 330 to a 4-f SG rate and filtered by a bandpass filter 332, which produces a 4f SG rate sequence of samples occupying the passband of bandpass filter 332. The sequencer 324, interpolator 330 and bandpass filter 332 may be implemented in the same manner as the like elements in FIGURE 11, using the configuration shown in FIGURE 14.

The 4f G rate comb filtered luminance signal sequence is applied to one input of an adder 518 .

-^J EA OMPI The original composite video signal samples that are applied to the bandpass filter 312 are also applied to a lowpass filter 513, which is clocked by the 4F 9G clock signal. The lowpass filter 513 has a complementary ^'5 frequency response to that of the bandpass filter 312.

That is, the transition bands of the two filters are equal but opposite, intersecting approximately at the 6dB point. The lowpass filtered samples are delayed by delay line 17 to bring the samples into time coincidence with 0 corresponding samples produced by the bandpass filter 332. The delayed samples which occupy the low frequency portion ' of the spectrum, are additively combined with high frequency, comb filtered luminance signals from bandpass filter 332 by adder 518, thereby producing a luminance 5 signal which is comb filtered over that portion of the spectrum formerly occupied by the chrominance information.

A digital lowpass filter, suitable for use as lowpass filter 13 in FIGURE 18, is shown in detail in FIGURE 19. The composite video signal samples are applied 0 to the first stage τ, of a twenty-two stage shift register 540. The input of the first stage and the outputs of all the stages are coupled to respective coefficient weighting circuits 550. Signals tapped from the shift register 540 are weighted by the coefficient weighting circuits and the 5 tap-weighted signals are combined by an adder tree 560.

The adder tree 560 includes delays shown at t_2A". t,." and t_4A" exhibiting delays which are multiples of the propagation delays exhibited by the adders in the tree. These delays equalize the signal delays in each of the 0 paths between the coefficient weighting circuits 550 and the output of the final adder 561 of the adder tree. Lowpass filtered samples at the output of adder 561 are latched into an output latch L_T -Ui_ by the 4FS_C_ clock signal.

The lowpass filter of FIGURE 19 exhibits a total 5 propagation delay equal to the time required for signals to be shifted from the input of the first stage τ₁₁"'to the output of the center stage τ,,¹", plus the propagation time of a coefficient weighting circuit, plus the propagation delay of six adders. If the coefficient weighting circuits are assumed to exhibit a propagation delay of 66 nanoseconds and the adders are assumed to provide a delay of 40 nanoseconds each, the time interval ^'5 required for a signal to appear at the output of adder 561 is approximately 15.37 cycles of the 4F SG clock.

Therefore, filtered samples will appear at the output of latch L_τ L_τP, sixteen 4Fsc_., clock cy ^■*cles after the corresponding samples are applied to the input of the 0 filter. This delay of the lowpass filter decreases the delay required for delay line 317 to bring the filtered samples into time correspondence with samples produced by the bandpass filter 332. For instance, if comb filtered samples are delayed by forty-three 4F 9G clock cycles 5 between the input of the bandpass filter 312 and the output of the bandpass filter 332, which was the case in the embodiment of FIGURES 12 and 14, this delay could be matched by the sixteen cycles of delay of the lowpass filter plus the delay of twenty-seven stages of the shift 0 register of the input bandpass filter 312. In such a configuration, the input of the first stage t,^IM of the lowpass filter of FIGURE 19 would be coupled to the output of the twenty-seventh stage of the bandpass filter shift register (FIGURE 12), thereby eliminating the need for a 5 separate delay line 317.

Several of the concepts inherent in the present invention are graphically illustrated by the idealized waveforms of FIGURE 20. The A/D converter produces video signal samples occupying the conventional video band, such 0 as that shown in FIGURE 20 for an NTSC color signal. The video signal includes luminance (Y) information extending from zero to about 4.2 MHz. I color mixture signal information is interleaved with the luminance information from about 2.1 to 4.2 MHz, and Q color mixture signal 5 information is interleaved over the range from about 3 to 4.2 MHz. In the range from 3 to 4.2 MHz, the color mixture signal information is modulated as double sideband signals of the color subcarrier f 9 C (3.58 MHz). When the composite video signal is filtered by bandpass filters 312, a passband as shown in FIGURE 20b results, containing the color mixture signal information between high and low cutoff frequencies f_, and f... By reason of the subsampling of the bandpass filtered signals at phases φ, and φ₂ of an f rate subsampling signal, replicas of the passband of FIGURE 20b are reproduced about the f subsampling frequency and harmonics thereof, several of which are shown in FIGURE 12c. The lowest frequency replicated band 650 includes demodulated I and Q color mixture signal information by reason of the f G subcarrier subsampling frequency. The next passband 652 comprises two passbands which overlap or alias between 3 and 4.2 MHz about the f 9G frequency. However, by reason of the phase and frequency relationships of the signal components in the overlap region, signal components reinforce one another. No interference components destructive of the color information signals are created by this aliasing. A portion of a third, higher frequency passband 654 is also shown.

The subsampled signals are then comb filtered, which effectively comb filters all of the replicated passbands, as illustrated by FIGURE 20d. The passband 650' now contains comb filtered and demodulated color mixture signal information. The "teeth" of the comb filter responses are separated by 15,734 Hz intervals, the horizontal line rate, for the NTSC signal.

The output bandpass filter passes that portion of the bands of FIGURE 20d which are located between cutoff frequencies f.. and f_H. This band contains modulated and comb filtered chrominance information. When the signal of this passband are subtracted from the composite signal of FIGURE 20a, a complementary comb filtered luminance signal is produced as shown in FIGURE 20e, which is comb filtered over the f_r Li to f--ti, frequency band. The output bandpass filter 332 is required to eliminate the signals in the passband 650' of FIGURE 20d

- T i OMP from the chrominance signals which are subtracted from the composite video signal. The combination of signals in passband 650' with the composite video signal, if allowed to occur, would undesirably introduce unwanted signal information into the low frequency region of the luminance signal.

- O EX OMPI

Claims

WHAT IS CLAIMED IS:

1. In a television signal processing system, including a source of sampled data video signals at a given data rate having first and second interleaved signal components in a passband, a signal separation system comprising: means, having an input coupled to said signal source, and an output, for passing said sampled data video signals at a reduced data rate which is less than said given data rate over a portion of said passband; and a comb filter, responsive to the output of said reduced data rate signal passing means, and operated at said reduced data rate for producing a comb filtered output signal.

2. In a television signal processing system, including a source of sampled data video signals at a given data rate having first and second interleaved signal components in a passband which are to be separated, a signal separation system comprising: means, having an input coupled to said signal source, and an output, for passing said sampled data video signals at a reduced rate which is less than said given data rate over a portion of said passband; a comb filter, having an input coupled to the output of said reduced data rate signal passing means, and operated at said reduced data rate for producing a first comb filtered output signal at said reduced data rate; an interpolator having an input coupled to receive said first comb filtered signal and an output at which a replica of said first comb filtered signal is produced at said given data rate; and means, having a first input coupled to said signal source to receive said sampled data video signals, a second input coupled to receive said replica of said first comb filtered signal, and an output, for combining said received signals to produce a second comb filtered signal.

3. The system of Claim 1 or 2, wherein said reduced data rate signal passing means comprises: ^'5 means for subsampling said video signals at a reduced rate which is less than said given data rate; and a bandpass filter responsive to said subsampled video signals for passing reduced rate filtered signal samples occupying a portion of said passband.

0 4. The system of Claim 2, wherein said first and second interleaved signal components comprise chrominance and luminance information signals; said first comb filtered output signal comprises a comb filtered chrominance signal; and wherein said combining means 5 comprises means for subtractively combining said sampled data video signals and said comb filtered chrominance signal to produce a comb filtered luminance signal.

5. A system for separating by comb filtering sampled data signals occurring at a given rate and 0 including first and second frequency interleaved signal components comprising: means for selecting prescribed ones of said sampled data signals and producing a signal having a rate reduced from said given rate, said reduced rate satisfying 5 the Nyquist criterion for sampled signals with respect to at least one of said first and second components; a comb filter operating at said reduced rate and responsive to said selected samples for producing a first combed manifestation of said selected samples; 0 an interpolator responsive to said first comb filtered signal for producing a comb filtered signal at said given rate; and means for combining the combed filtered signal from the interpolator and the sampled data signals to 35 produce a second comb filtered signal, wherein said first and second comb filtered signals correspond substantially to said first and second signal components, respectively.

6. In a television signal processing system, including a source of sampled data video signals at a given data rate having first and second frequency interleaved signal components which are to be separated, wherein one video line interval comprises a given number of signal samples; a signal separation system comprising: means, having an input coupled to said signal source, and an output for producing sample data video signals, in response to a subsampling signal, which include said first and second interleaved signal components at a reduced rate which is less than said given rate, with one video line interval comprising a reduced number of signal samples which is not an integer submultiple of said given number; a comb filter, including a delay line, and responsive to said reduced rate sampled data,video signals for producing a comb filtered output signal; and means for changing the phase of said subsampling signal from one video line interval to another.

7. In a television receiver, including a source of analog video signals occupying a given passband and having first and second frequency interleaved signal components, a signal separation system comprising: an analog to digital converter, responsive to a sampling clock signal, for producing a sequence of digital video signal samples in response to said sampling clock signal; a bandpass filter, having an input responsive to said sequence of digital video signal samples for sampling said sequence at first and second phases of a subsampling clock signal and producing filtered signal samples at a reduced rate relative to said input sequence which occupy a portion of said given passband;

OMPI a comb filter including a shift register and a first signal combining circuit responsive to said reduced rate filtered signal samples, said first signal combining circuit being further responsive to delayed filtered signal samples provided by said shift register for producing a first comb filtered signal sequence at said reduced rate at an output which corresponds to said first frequency interleaved signal components; an interpolator having an input coupled to the output of said comb filter and responsive to said first comb filtered signal sequence for generating interpolated signals therefrom and producing a replica of said first comb filtered signal sequence at the rate of said input sequence; and a second signal combining circuit having a first input responsive to said replica of said first comb filtered signal sequence, a second input responsive to said sequence of digital video signal samples produced by said analog to digital converter, and an output at which a second comb filtered signal sequence corresponding to said second interleaved signal components is produced.

8. The system of Claim 7, further comprising: a second shift register, coupled between said analog to digital converter and said second input of said second signal combining circuit for providing an equalizing delay to said sequence of digital video signal samples produced by said analog to digital converter.

9. The system of Claim 7 or 8, wherein said sampling clock signal and said subsampling clock signal are aligned in phase with the color subcarrier of said video signals, said first signal combining circuit is a subtractive signal combining circuit which produces a comb filtered chrominance signal sequence, and said second signal combining circuit is a subtractive signal combining circuit which produces a comb filtered luminance signal sequence.

10. The system of Claim 7 or 8 further comprising: means for producing a sampling clock signal containing sampling pulses produced at first, second, -5 third and fourth phases of the color subcarrier signal component of said video signals; wherein said analog to digital converter is responsive to said sampling clock signal; and means for producing first and second subsampling 0 clock signals containing subsampling pulses corresponding to said first and second phases, wherein said bandpass filter is responsive to said subsampling clock signals.

11. The system of Claim 10, further comprising 5 means for producing a third subsampling clock signal at a reduced rate relative to said sampling clock signal and in phase alignment therewith, wherein said comb filter is responsive to said third subsampling clock signal.

0 12. The system of Claim 10, wherein said means- for producing said first and second subsampling clock signals further includes means for producing third and fourth subsampling clock signals containing subsampling pulses corresponding to said third and fourth phases, 5 wherein said bandpass filter is responsive to said first and second subsampling clock signals during a first video line, and to said third and fourth subsampling clock signals during a second, succeeding video line.

13. In a television signal processing system, 0 including a source of sampled data video signals at a given data rate having first and second frequency interleaved signal components in a video passband, a signal separation system comprising: a first bandpass filter having an input 5 responsive to said sampled data video signals and an output at which filtered signal samples occupying a portion of said video passband are produced at said given data rate; a comb filter, responsive to said filtered signal samples, and operating at a reduced data rate which is less than said given data rate, for producing a comb filtered signal at said reduced data rate; and a second bandpass filter responsive to said comb filtered signal for producing a comb filtered signal at said given data rate which occupies a portion of said video passband.

14. In a television signal processing system, including a source of sampled data video signals at a given data rate having first and second frequency interleaved signal components in a video passband, a signal separation system comprising: a first bandpass filter having an input responsive to said sampled data video signals and an output at which filtered signal samples occupying a portion of said video passband are produced at said given data rate; a comb filter, responsive to said filtered signal samples, and operating at a reduced data rate which is less than said given data rate, for producing a comb filtered signal at said reduced data rate; an interpolator having an input responsive to said comb filtered signal and an output at which a replica of said comb filtered signal is produced at said given data rate; and a second bandpass filter, coupled to the output of said interpolator, and responsive to said comb filtered signal at said given data rate for producing a comb filtered signal which occupies a portion of said video passband.

OMPI WIPO .

15. The system of Claim 13 or 14, further comprising: means, responsive to said comb filtered signal produced by said second bandpass filter and to said video signals provided by said signal source, for combining said signals to produce a second comb filtered signal.

16. In a television receiver, including a source of analog video signals occupying a given passband and containing first and second frequency interleaved signal components, a signal separation system comprising: an analog to digital converter responsive to said analog video signals for producing a sequence of digitized video signal samples in response to a sampling signal of a given frequency; a first bandpass filter having an input responsive to said digitized video signal sequence ^*and an output at which a sequence of filtered video signal samples occupying a portion of said passband is produced at said given frequency; a first comb filter having an input responsive to selected ones of said filtered video signal samples and an output at which a first sequence of comb filtered signal samples is produced in a first phase relationship with said filtered video signal sample sequence; a second comb filter having an input responsive to selected ones of said filtered video signal samples and an output at which a second sequence of comb filtered signal samples is produced in a second phase relationship with said filtered video signal sample sequence; means for combining said first and second sequences of comb filtered signal samples to produce a first comb filtered signal corresponding to said first interleaved signal components; a second bandpass filter having an input responsive to said first comb filtered signal for producing a replica of said first comb filtered signal in

O FI /fa WIPO a sequence of signal samples at said given frequency which occupies a portion of said passband; and a signal combining circuit having a first input responsive to said replica of said first comb filtered -5 signal, a second input responsive to said sequence of digitized video signal samples produced by said analog to digital converter, and an output at which a second comb filtered signal corresponding to said second interleaved signal components is produced.

0 17. The system of Claim 16, further comprising: means for producing a sampling clock signal including sampling pulses of first, second, third and fourth phases, wherein said analog to digital converter and 5 said bandpass filters are responsive to said sampling clock signal; means for producing a first subsampling clock signal including subsampling pulses corresponding to the first phase of said sampling clock signal, 0 wherein said first comb filter is responsive to said first subsampling clock signal for selecting said ones of said filtered video signal samples; and means for producing a second subsampling clock signal including subsampling pulses corresponding to a 5 second phase of said sampling clock signal, wherein said second comb filter is responsive to said second subsampling clock signal for selecting said second-named ones of said filtered video signal samples.

18. The system of Claim 17, further including: 0 means for producing a third subsampling clock signal including subsampling pulses corresponding to a third phase of said sampling clock signal; and means for producing a fourth subsampling clock signal including subsampling pulses corresponding to a 35 fourth phase of said sampling clock signal;

O PI

^' wherein said comb filters are responsive to said first and second subsampling clock signals during a first video line, and to said third and fourth subsampling clock signals during a second, succeeding video line.

19. The system of Claim 16 or 18, wherein said first comb filtered signal comprises a comb filtered chrominance signal; and wherein said signal combining circuit comprises a subtractive signal combining circuit responsive to a replica of said comb filtered chrominance signal and said sequence of digitized video signal samples for producing a comb filtered luminance signal.

20. The system of Claim 16, wherein said first comb filtered signal exhibits a sample rate which is less than said given frequency, and said second bandpass filter is responsive to said first comb filtered signal for interpolating sample values so as to produce said replica of said first comb filtered signal exhibiting a sample rate which is equal to said given frequency.

21. The system of Claim 16 or 20, wherein said- first and second bandpass filters exhibit substantially the same passband response characterisitic; and further comprising: an equalizing delay coupled between said analog to digital converter and said second input of said signal combining circuit.

22. A signal separation system for comb filtering a sampled data composite signal of a given data rate including first and second frequency interleaved components comprising: a first comb filter responsive to every Nth signal sample for generating a first comb filtered signal where N is an integer; a second comb filter responsive to every Nth signal sample different from said samples in said first

-^JRE

OMPI

/fa_^ WIPO comb filter for generating a second comb filtered signal, said first and second comb filters operating at a lesser rate than the rate of said sampled data composite signal; means responsive to said first and second comb filtered signals for generating a third comb filtered signal at said sampled data composite signal rate, said third comb filtered signal corresponding to a combed manifestation of one of said first and second interleaved components; a bandpass filter responsive to said third comb filtered signal for passing a prescribed spectrum thereof; and means responsive to said sampled data composite signal and the signal produced by said bandpass filter for producing a fourth comb filtered signal wherein said fourth comb filtered signal and said third further signal correspond substantially exclusively to said first and said second frequency interleaved signal components respectively.

23. In a television signal processing system, including a source of sampled data video signals at a given data rate having frequency interleaved luminance and chrominance signal components in a video passband, a signal separation system comprising: a first bandpass filter having an input responsive to said sampled data video signals for producing a sequence of filtered signal samples occupying a portion of said passband; a comb filter, responsive to selected ones of said filtered signal samples for producing a comb filtered luminance signal at a reduced rate which is less than said given data rate; a second bandpass filter having an input responsive to said comb filtered luminance signal and an output at which a comb filtered luminance signal is produced at said given data rate which occupies a portion of said passband; and fu:

OMPI

WIPO means, having a first input responsive to said comb filtered luminance signal produced by said second bandpass filter, a second input responsive to said sampled _. data video signals which occupy a portion of said video ^' 5 passband, and an output for producing a comb filtered luminance signal occupying said video passband.

24. The system of Claim 23, further comprisng: a lowpass filter coupled between said source of sampled data video signals and said second input of said 0 comb filtered luminance signal producing means, said lowpass filter exhibiting a complementary passband response characteristic to that of said second bandpass filter.

25. The system of Claim 23 or 24, wherein said 5 comb filter comprises a shift register, an additive combining circuit, and a subtractive combining circuit, each having an input responsive to selected ones of said filtered signal samples, the output of said shift register being coupled to second inputs of said additive and 0 subtractive combining circuits, said additive combining circuit producing said reduced data comb filtered luminance signal, and said subtractive combining circuit producing a reduced rate comb filtered chrominance signal.